Panel source and identity verification stage
1. Matching IP and registered address: The regional restriction of online questionnaire surveys determines that there are more questionnaires in large cities such as North, Shanghai, and Guangzhou than in third or fourth-tier small towns.The matching of IP and address eliminates the possibility of users filling in registered addresses maliciously in order to get more questionnaires. The matching of IP and address prevents users from maliciously filling in the registered address in order to get more questionnaires.
2. Limit and remove multiple registrations: The same IP and device (including computers and cell phones) can only be registered once. Especially nowadays with mobile devices, we can lock accounts by device number and multiple accounts will be removed.
3, machine registration auto-detection: this is not much to say, the first line of defense for security.
4, Alipay real name authentication, cell phone number unique, bound WeChat identity verification: through Alipay, WeChat recruitment, using Alipay real name and WeChat unique third-party way to ensure the uniqueness and authenticity of the account. Ensure that every opinion comes from real respondents.
5, ID number filling and uploading ID photo of the ultimate verification method.
Access Stage
6, Device Uniqueness: whether it's a computer or a cell phone, each device can be limited to uniqueness by cookie and device number.
7, IP check: the same IP can only answer once, while not the questionnaire target city of the respondent IP restrictions on answering the questionnaire. Suitable for questionnaires with strong geographical restrictions. For example, if we implement the Xiamen local beer program, non-Xiamen IP respondents can't enter the questionnaire even if they receive an invitation to participate in the questionnaire.
8. Matching of demographic attributes: The demographic attributes filled in during registration and the information in the questionnaire are double-checked. Any inconsistency will be ejected.
9, the length of the answer to the test: the answer time is too short is likely to cause not seriously look at the question. And some, such as the concept test, advertising test items answer time is too long, it is likely to have forgotten the previous test content.
10, straight-line question detection and scoring questions split: for the questionnaire scoring questions more cases, in the process of programming on the realization of the random split, each page only for up to 5 points option scoring. For all scores have been the straight line results are detected, prompted to re-select or jump out.
11, the set of trap questions: the situation is a lot of can be common sense questions to detect whether the respondent is seriously look at the question, can also be a test of logic. For example, if you choose to have purchased a car but the make and model of the car is not correct, it is likely that you are not the real owner of the car.
12, a fixed time within the answer to a unified theme of the questionnaire: according to the needs of the customer, excluding similar questionnaires.
13, clever screening questions set up: for example, the screening questions of the foreign wine program, we generally do not ask whether they have drunk foreign wine. Instead, we ask if you have consumed any of the following products. Only those who choose the foreign wine can continue to answer.
14, mark the key: different font colors to highlight the topic, so that the topic is easier to identify and understand. It is especially suitable for questionnaires that ask about multiple aspects of a product in a large number of repetitive categories.
Data cleansing and settlement
15. Checking and cleaning of open questions: Especially for questionnaires with a lot of open questions, there are a lot of lazy people who have cancer, and the more open questions, not only do they quit in the middle of the questionnaires, but they may also be answered inappropriately and copied and pasted. On the one hand, we suggest reducing the number of open-ended questions and on the other hand, we suggest cleaning up the data.
16, logic check: for the logic of the program there is no way to control the logic of the error, we are in the data processing in the abolition of the volume.
17, data format checking: in the programming stage we will be for the date, cell phone number and other specific data format to provide respondents to fill in the answer, data cleaning stage again for the difficult to control the data for a second check.
Last but not least
18, material and non-material incentives coexist, the right people and questionnaires to give the right reward points: in the daily panel maintenance, we give members a reasonable reward points for each questionnaire, for example, for high-end car owners will be higher than the general online shopping questionnaire incentives. At the same time, we give members non-material incentives other than material incentives, such as member badges, privileges and other non-material incentives.
There are a lot of questionnaire research platforms on the internet, you can use Le Survey to design questionnaires for free
.